Lab

Tools built to scale creative
quality across teams.

These aren't side projects or experiments. They're professional infrastructure — tools I designed and built because a gap existed between what the organization needed to create and what the design team could produce alone.

Each one started with a specific problem: a team spending hours on a task that should take minutes, or producing materials that didn't match the brand because they didn't have the tools to do otherwise. The solutions use AI as a layer of creative intelligence on top of defined systems — not to replace design judgment, but to extend it to people who don't have it by default.

Problem: Non-designers creating off-brand materials

Brand-Compliant Material Generator

[PLACEHOLDER: Describe what this tool does in concrete terms.

Prompts to answer:
— Who uses this tool and what were they doing before it existed?
— What inputs do they provide? What does the tool produce?
— How does it enforce brand compliance — what guardrails did you build in?
— What's an example of something someone made with it that they couldn't have made before?]

Claude API Brand Systems [PLACEHOLDER: other tech]

[PLACEHOLDER: Impact — e.g. "Reduced turnaround time for sales collateral from X days to X hours. Now used by [N] non-designers across marketing, sales, and product."]

[PLACEHOLDER: Tool screenshot or interface mockup]

[PLACEHOLDER: What problem does this tool solve?]

[PLACEHOLDER: Second AI Tool Name]

[PLACEHOLDER: Describe the tool.

Prompts to answer:
— What workflow did this replace or augment?
— Who is the user — designer, non-designer, executive?
— What does the AI layer specifically do? What does the human still control?
— What makes this a design problem, not just a technical one?]

Claude API [PLACEHOLDER: tech tags]

[PLACEHOLDER: Impact — specific metric or qualitative outcome]

[PLACEHOLDER: Tool screenshot or demo]

[PLACEHOLDER: Problem statement]

[PLACEHOLDER: Third Tool — or remove this block]

[PLACEHOLDER: If you have a third tool, describe it here. If not, remove this lab-item block entirely.]

[PLACEHOLDER: tech tags]

[PLACEHOLDER: Impact]

[PLACEHOLDER: Visual]

"The most interesting design question in AI isn't 'what can the model do' — it's 'what system do you need to build around the model so that the output is consistently, reliably on-brand?'"

Joe Howard  â€”  on AI-assisted creative workflow

How these get built

Every tool here started with a conversation — usually something like "this takes us three days and it shouldn't." I map the workflow, identify where human judgment is genuinely required versus where consistent rules would produce better outcomes than inconsistent intuition, and build the AI layer to own the latter.

The design work isn't in the prompting. It's in the system architecture: what constraints you define, what variables you expose, what the user controls, and what the output looks like. Getting that right is a design problem that happens to require knowing how language models work.